Recurrent pulmonary symptoms in immunocompromised (IC) patients are a common problem that poses diagnostic and therapeutic challenges. Repeat fiberoptic bronchoscopy (FOB) is often performed because it is relatively safe and more invasive procedures may be contraindicated. This study's purpose was to evaluate the diagnostic and therapeutic efficacies of repeat FOB in IC patients. A retrospective chart review was performed for all IC patients undergoing repeat FOB in our department from January 1, 1987 through June 30, 1992. Patients were classified as having either "persistent" pulmonary disease (as documented by repeat FOB performed within 30 days of initial FOB) or "new" disease (as documented by repeat FOB performed after 30 days from the initial FOB). In patients with HIV infection, a new diagnosis was found in 8 of 25 (32%) repeat FOBs with new disease and 2 of 20 (10%) with persistent disease (P=NS). A change in therapy was initiated in 14 of 25 (56%) repeat FOB with new disease, and 7 of 20 (35%) with persistent disease (P=NS). In hematologic/oncologic disorders, a new diagnosis was found in 10 of 20 (50%) repeat FOBs with new disease and in 12 of 42 (29%) with persistent disease (P=0. 10); a change in therapy was initiated in 14 of 20 (70%) repeat FOBs with new disease and in 24 of 42 (57%) with persistent disease (P=NS). In patients with other immunocompromised conditions, a new diagnosis was found in 3 of 10 (30%) repeat FOB with new disease and 1 of 4 (25%) with persistent disease (P=NS). A change in therapy was initiated in 5 of 10 (50%) repeat FOBs with new disease and in 2 of 4 (50%) with persistent disease. We concluded that the diagnostic yield for repeat FOB in patients with HIV infection was equally effective in the presence of new or persistent disease. In patients with hematologic/oncologic disorders and new disease, repeat FOB may be more likely to establish a new diagnosis (P=0.10).